Mapping LA Traffic Accidents¶

In [1]:
#We'll start by importing the libraries we need
# for data wrangling
import pandas as pd

# for interactive plots
import plotly.express as px

# to import open data
from sodapy import Socrata
In [2]:
# make sure to install these packages before running:
# pip install pandas
# pip install sodapy

import pandas as pd
from sodapy import Socrata

# Unauthenticated client only works with public data sets. Note 'None'
# in place of application token, and no username or password:
client = Socrata("data.lacity.org", None)

# First 2000 results, returned as JSON from API / converted to Python list of
# dictionaries by sodapy.
results = client.get("d5tf-ez2w", limit=2000)

# Convert to pandas DataFrame
collisions = pd.DataFrame.from_records(results)

# print it with .sample, which gives you random rows
collisions.sample(2)

WARNING:root:Requests made without an app_token will be subject to strict throttling limits.
Out[2]:
dr_no date_rptd date_occ time_occ area area_name rpt_dist_no crm_cd crm_cd_desc mocodes ... premis_desc location cross_street location_1 :@computed_region_qz3q_ghft :@computed_region_k96s_3jcv :@computed_region_tatf_ua23 :@computed_region_ur2y_g4cx :@computed_region_kqwf_mjcx :@computed_region_2dna_qi2s
1806 191113760 2019-07-19T00:00:00.000 2019-07-19T00:00:00.000 0955 11 Northeast 1142 997 TRAFFIC COLLISION 3101 3401 3701 3008 3026 ... STREET HILLHURST AV AMBROSE AV {'latitude': '34.11', 'longitude': '-118.2874'... 23445 400 511 NaN 7 3
764 181106721 2018-03-02T00:00:00.000 2018-03-02T00:00:00.000 2025 11 Northeast 1149 997 TRAFFIC COLLISION 3101 3401 3701 3003 3024 3602 ... STREET FIGUEROA ST AVENUE 54 {'latitude': '34.1076', 'longitude': '-118.196... 23673 374 588 28 11 93

2 rows × 24 columns

In [3]:
# Convert to pandas DataFrame
collisions = pd.DataFrame.from_records(results)

Data Exploration and Analysis¶

In [4]:
# how many rows and columns?
collisions.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 2000 entries, 0 to 1999
Data columns (total 24 columns):
 #   Column                       Non-Null Count  Dtype 
---  ------                       --------------  ----- 
 0   dr_no                        2000 non-null   object
 1   date_rptd                    2000 non-null   object
 2   date_occ                     2000 non-null   object
 3   time_occ                     2000 non-null   object
 4   area                         2000 non-null   object
 5   area_name                    2000 non-null   object
 6   rpt_dist_no                  2000 non-null   object
 7   crm_cd                       2000 non-null   object
 8   crm_cd_desc                  2000 non-null   object
 9   mocodes                      1947 non-null   object
 10  vict_age                     1879 non-null   object
 11  vict_sex                     1958 non-null   object
 12  vict_descent                 1947 non-null   object
 13  premis_cd                    1958 non-null   object
 14  premis_desc                  1958 non-null   object
 15  location                     2000 non-null   object
 16  cross_street                 1851 non-null   object
 17  location_1                   2000 non-null   object
 18  :@computed_region_qz3q_ghft  1999 non-null   object
 19  :@computed_region_k96s_3jcv  1978 non-null   object
 20  :@computed_region_tatf_ua23  1993 non-null   object
 21  :@computed_region_ur2y_g4cx  750 non-null    object
 22  :@computed_region_kqwf_mjcx  1979 non-null   object
 23  :@computed_region_2dna_qi2s  1886 non-null   object
dtypes: object(24)
memory usage: 375.1+ KB
In [5]:
# what fields and datatypes?
results = collisions.dtypes

print(results)

dr_no                          object
date_rptd                      object
date_occ                       object
time_occ                       object
area                           object
area_name                      object
rpt_dist_no                    object
crm_cd                         object
crm_cd_desc                    object
mocodes                        object
vict_age                       object
vict_sex                       object
vict_descent                   object
premis_cd                      object
premis_desc                    object
location                       object
cross_street                   object
location_1                     object
:@computed_region_qz3q_ghft    object
:@computed_region_k96s_3jcv    object
:@computed_region_tatf_ua23    object
:@computed_region_ur2y_g4cx    object
:@computed_region_kqwf_mjcx    object
:@computed_region_2dna_qi2s    object
dtype: object
In [6]:
# what are the first 5 rows?
collisions.head(5)
Out[6]:
dr_no date_rptd date_occ time_occ area area_name rpt_dist_no crm_cd crm_cd_desc mocodes ... premis_desc location cross_street location_1 :@computed_region_qz3q_ghft :@computed_region_k96s_3jcv :@computed_region_tatf_ua23 :@computed_region_ur2y_g4cx :@computed_region_kqwf_mjcx :@computed_region_2dna_qi2s
0 190319651 2019-08-24T00:00:00.000 2019-08-24T00:00:00.000 0450 03 Southwest 0356 997 TRAFFIC COLLISION 3036 3004 3026 3101 4003 ... STREET JEFFERSON BL NORMANDIE AV {'latitude': '34.0255', 'longitude': '-118.300... 22724 691 916 7 14 32
1 190319680 2019-08-30T00:00:00.000 2019-08-30T00:00:00.000 2320 03 Southwest 0355 997 TRAFFIC COLLISION 3037 3006 3028 3030 3039 3101 4003 ... STREET JEFFERSON BL W WESTERN {'latitude': '34.0256', 'longitude': '-118.308... 23079 686 919 7 12 19
2 190413769 2019-08-25T00:00:00.000 2019-08-25T00:00:00.000 0545 04 Hollenbeck 0422 997 TRAFFIC COLLISION 3101 3401 3701 3006 3030 ... STREET N BROADWAY W EASTLAKE AV {'latitude': '34.0738', 'longitude': '-118.207... 23448 488 516 NaN 11 1
3 190127578 2019-11-20T00:00:00.000 2019-11-20T00:00:00.000 0350 01 Central 0128 997 TRAFFIC COLLISION 0605 3101 3401 3701 3011 3034 ... STREET 1ST CENTRAL {'latitude': '34.0492', 'longitude': '-118.239... 22728 536 1502 NaN 9 28
4 190319695 2019-08-30T00:00:00.000 2019-08-30T00:00:00.000 2100 03 Southwest 0374 997 TRAFFIC COLLISION 0605 4025 3037 3004 3025 3101 ... STREET MARTIN LUTHER KING JR ARLINGTON AV {'latitude': '34.0108', 'longitude': '-118.318... 22725 747 1017 7 14 35

5 rows × 24 columns

Looking at 2022 Traffic Collisions¶

In [7]:
#only look at traffic collisions from January 2022 to December 2022
year_results = client.get("d5tf-ez2w", 
                     limit = 50000, # putting an arbitrary high number (otherwise defaults to 1000)
                     where = "date_occ between '2022-01-01T00:00:00' and '2022-12-31T00:00:00'"
                    )
In [8]:
# Convert to pandas DataFrame
lastyear_collisions = pd.DataFrame.from_records(year_results)
In [9]:
#check the data
# how many rows and columns?
lastyear_collisions.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 16354 entries, 0 to 16353
Data columns (total 18 columns):
 #   Column        Non-Null Count  Dtype 
---  ------        --------------  ----- 
 0   dr_no         16354 non-null  object
 1   date_rptd     16354 non-null  object
 2   date_occ      16354 non-null  object
 3   time_occ      16354 non-null  object
 4   area          16354 non-null  object
 5   area_name     16354 non-null  object
 6   rpt_dist_no   16354 non-null  object
 7   crm_cd        16354 non-null  object
 8   crm_cd_desc   16354 non-null  object
 9   mocodes       16226 non-null  object
 10  vict_age      14826 non-null  object
 11  vict_sex      16043 non-null  object
 12  vict_descent  16026 non-null  object
 13  premis_cd     16354 non-null  object
 14  premis_desc   16354 non-null  object
 15  location      16354 non-null  object
 16  cross_street  15691 non-null  object
 17  location_1    16354 non-null  object
dtypes: object(18)
memory usage: 2.2+ MB
In [10]:
lastyear_collisions.tail(5)
Out[10]:
dr_no date_rptd date_occ time_occ area area_name rpt_dist_no crm_cd crm_cd_desc mocodes vict_age vict_sex vict_descent premis_cd premis_desc location cross_street location_1
16349 231204524 2023-01-10T00:00:00.000 2022-12-31T00:00:00.000 1521 12 77th Street 1269 997 TRAFFIC COLLISION 4025 3004 3028 3030 37 M H 101 STREET 300 E MANCHESTER AV NaN {'latitude': '33.9638', 'longitude': '-118.262...
16350 230204688 2022-12-31T00:00:00.000 2022-12-31T00:00:00.000 1240 02 Rampart 0218 997 TRAFFIC COLLISION 0605 3030 4024 35 M O 101 STREET 1500 W SUNSET BL NaN {'latitude': '34.0758', 'longitude': '-118.255...
16351 230904001 2023-01-01T00:00:00.000 2022-12-31T00:00:00.000 2225 09 Van Nuys 0964 997 TRAFFIC COLLISION 3038 3602 3101 3004 3026 4026 3029 3037 58 M W 101 STREET WOODMAN AV ADDISON ST {'latitude': '34.1612', 'longitude': '-118.431...
16352 222019939 2022-12-31T00:00:00.000 2022-12-31T00:00:00.000 1240 20 Olympic 2076 997 TRAFFIC COLLISION 3006 3028 3030 4027 3034 3037 3101 3401 3701 22 NaN NaN 101 STREET VERMONT AV WASHINGTON BL {'latitude': '34.04', 'longitude': '-118.2916'...
16353 230106871 2022-12-31T00:00:00.000 2022-12-31T00:00:00.000 1225 01 Central 0162 997 TRAFFIC COLLISION 0605 3030 4024 47 M W 101 STREET W 8TH ST S OLIVE ST {'latitude': '34.0452', 'longitude': '-118.256...

Data cleanup (asssisted by Yoh)¶

Here we split the data in location_1 column that is in a json (dictionary) format into separate columns for latitude and longitude.

This allows the map further below to work!

In [11]:
# create latitudue/longitude columns
lastyear_collisions['latitude'] = pd.json_normalize(lastyear_collisions['location_1'])['latitude']
lastyear_collisions['longitude'] = pd.json_normalize(lastyear_collisions['location_1'])['longitude']

# convert to floats
lastyear_collisions['latitude'] = lastyear_collisions['latitude'].astype('float')
lastyear_collisions['longitude'] = lastyear_collisions['longitude'].astype('float')

Mapping Traffic Collision Data 2022¶

Allowing us to see the traffic collisions in an interactive map paints a picture of the traffic violence that is occuring in LA.

In [12]:
fig = px.scatter_mapbox(lastyear_collisions,
                        lat='latitude',
                        lon='longitude',
                        mapbox_style="carto-positron")

# options on the layout
fig.update_layout(
    width = 800,
    height = 800,
    title = "Interactive LA Traffic Collisions Map (2022)",
    title_x=0.5 # aligns title to center
)

fig.show()

Now let's map traffic collisions by race¶

mapping collisions by race is a good way to see the demographics of the accidents and seeing the patterns of where certain groups live and how they commute. In future steps, we can then mark where Vision Zero initiatives are being implemented and see if communities with high rates of collisions are being targeted or not.

In [13]:
fig = px.scatter_mapbox(lastyear_collisions, 
                        lat="latitude", 
                        lon="longitude", 
                        color="vict_descent",
                        labels={'vict_descent':'Race'},
                        mapbox_style="carto-positron"
                       )

# options on the layout
fig.update_layout(
    width = 800,
    height = 800,
    title = "Interactive LA Traffic Collisions Map (2022)",
    title_x=0.5 # aligns title to center
)


fig.show()

Race is as follows:¶

A - Other Asian B - Black C - Chinese D - Cambodian F - Filipino G - Guamanian H - Hispanic/Latin/Mexican I - American Indian/Alaskan Native J - Japanese K - Korean L - Laotian O - Other P - Pacific Islander S - Samoan U - Hawaiian V - Vietnamese W - White X - Unknown Z - Asian Indian

In [14]:
#Creating collision maps isolating by race
def race_map(race='H'):
    
    fig = px.scatter_mapbox(
                            lastyear_collisions[lastyear_collisions.vict_descent==race], 
                            lat="latitude", 
                            lon="longitude", 
                            color="vict_descent",
                            labels={'vict_descent':'Race'},
                            mapbox_style="carto-positron"
    )

    # options on the layout
    fig.update_layout(
        width = 600,
        height = 600,
        title = "Interactive LA Traffic Collisions Map (2022)<br>Showing where descent label is "+race,
        title_x=0.5 # aligns title to center
    )

    fig.show()
In [15]:
#showing collision map involving only those labeled as Hispanic/Latin/Mexican
race_map(race='H')

getting rid of null values¶

In [16]:
# get a list of distinct race values
race_list = lastyear_collisions.vict_descent.unique().tolist()
race_list
Out[16]:
['H',
 'O',
 'W',
 'F',
 nan,
 'A',
 'B',
 'V',
 'X',
 'K',
 'U',
 'C',
 'P',
 'Z',
 'J',
 'S',
 'G']
In [17]:
#dropping nan or null value
lastyear_collisions.dropna(inplace=True)
In [18]:
race_list = lastyear_collisions.vict_descent.unique().tolist()
race_list
Out[18]:
['H', 'O', 'W', 'F', 'A', 'B', 'X', 'K', 'U', 'V', 'C', 'P', 'Z', 'J', 'G']

Using a loop to see each race individually on the interactive map¶

By seeing each race individually, we can see where the accidents are happening to certain groups of people. From there, we can see if there are any discrepancies in those areas and if Vision Zero has done anything to improve those communities for traffic safety.

In [76]:
for race in race_list:
    race_map(race)

Summarize by Neighborhood¶

In [19]:
import geopandas as gpd
In [20]:
neighborhoods = gpd.read_file("https://services5.arcgis.com/7nsPwEMP38bSkCjy/arcgis/rest/services/LA_Times_Neighborhoods/FeatureServer/0/query?outFields=*&where=1%3D1&f=geojson")

ERROR 1: PROJ: proj_create_from_database: Open of /opt/conda/share/proj failed
In [21]:
neighborhoods.plot(figsize=(12,12))
Out[21]:
<AxesSubplot: >
In [22]:
# convert lastyear_collisions to geodataframe
gdf = gpd.GeoDataFrame(
    lastyear_collisions, geometry=gpd.points_from_xy(lastyear_collisions.longitude, lastyear_collisions.latitude),crs=4326)

# check that it can be mapped
gdf.plot()
Out[22]:
<AxesSubplot: >
In [23]:
# conduct a spatial join (i.e. assign each collision to the neighborhood it is in)
collision_by_neighborhood = gpd.sjoin(neighborhoods, gdf) #Spatial join Points to polygons
collision_by_neighborhood
Out[23]:
OBJECTID name geometry index_right dr_no date_rptd date_occ time_occ area area_name ... vict_age vict_sex vict_descent premis_cd premis_desc location cross_street location_1 latitude longitude
0 1 Adams-Normandie POLYGON ((-118.30069 34.03731, -118.30388 34.0... 3342 220307611 2022-03-17T00:00:00.000 2022-03-17T00:00:00.000 1230 03 Southwest ... 30 F H 101 STREET JEFFERSON BL HOBART BL {'latitude': '34.0256', 'longitude': '-118.307... 34.0256 -118.3078
0 1 Adams-Normandie POLYGON ((-118.30069 34.03731, -118.30388 34.0... 287 220304480 2022-01-12T00:00:00.000 2022-01-08T00:00:00.000 1900 03 Southwest ... 37 F B 101 STREET HOBART BL JEFFERSON BL {'latitude': '34.0256', 'longitude': '-118.307... 34.0256 -118.3078
0 1 Adams-Normandie POLYGON ((-118.30069 34.03731, -118.30388 34.0... 1415 220305437 2022-02-03T00:00:00.000 2022-02-03T00:00:00.000 1840 03 Southwest ... 63 M H 101 STREET JEFFERSON HOBART {'latitude': '34.0256', 'longitude': '-118.307... 34.0256 -118.3078
0 1 Adams-Normandie POLYGON ((-118.30069 34.03731, -118.30388 34.0... 4556 220309034 2022-04-13T00:00:00.000 2022-04-13T00:00:00.000 1815 03 Southwest ... 29 F H 101 STREET HARVARD 29TH {'latitude': '34.0291', 'longitude': '-118.306... 34.0291 -118.3067
0 1 Adams-Normandie POLYGON ((-118.30069 34.03731, -118.30388 34.0... 9130 220314071 2022-07-22T00:00:00.000 2022-07-20T00:00:00.000 2303 03 Southwest ... 21 M O 101 STREET HALLDALE AV 29TH ST {'latitude': '34.0291', 'longitude': '-118.302... 34.0291 -118.3028
... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...
113 114 Lake Balboa POLYGON ((-118.50116 34.22114, -118.50116 34.2... 5401 221008586 2022-04-30T00:00:00.000 2022-04-30T00:00:00.000 0930 10 West Valley ... 25 M H 101 STREET BURTON ENCINO {'latitude': '34.2201', 'longitude': '-118.514... 34.2201 -118.5143
113 114 Lake Balboa POLYGON ((-118.50116 34.22114, -118.50116 34.2... 14212 221016437 2022-11-13T00:00:00.000 2022-11-13T00:00:00.000 0745 10 West Valley ... 50 M O 101 STREET BALBOA BL BALBOA PL {'latitude': '34.2179', 'longitude': '-118.501... 34.2179 -118.5014
113 114 Lake Balboa POLYGON ((-118.50116 34.22114, -118.50116 34.2... 10487 221013002 2022-08-21T00:00:00.000 2022-08-20T00:00:00.000 2045 10 West Valley ... 67 F W 101 STREET BALBOA BL BALBOA PL {'latitude': '34.2179', 'longitude': '-118.501... 34.2179 -118.5014
113 114 Lake Balboa POLYGON ((-118.50116 34.22114, -118.50116 34.2... 10291 221712335 2022-08-15T00:00:00.000 2022-08-15T00:00:00.000 1133 17 Devonshire ... 19 M H 101 STREET ROSCOE BL LOUISE AV {'latitude': '34.2211', 'longitude': '-118.509... 34.2211 -118.5099
113 114 Lake Balboa POLYGON ((-118.50116 34.22114, -118.50116 34.2... 14445 221716408 2022-11-19T00:00:00.000 2022-11-18T00:00:00.000 1945 17 Devonshire ... 65 M O 101 STREET ROSCOE BL LOUISE AV {'latitude': '34.2211', 'longitude': '-118.509... 34.2211 -118.5099

13627 rows × 24 columns

In [24]:
# count the number of records per community
collision_counts = collision_by_neighborhood.value_counts('name').reset_index(name='collision_counts')
collision_counts
Out[24]:
name collision_counts
0 Downtown 811
1 Van Nuys 421
2 Hollywood 409
3 Westlake 391
4 Koreatown 345
... ... ...
109 Montecito Heights 14
110 Mount Washington 12
111 Rancho Park 10
112 Beverlywood 6
113 Chatsworth Reservoir 4

114 rows × 2 columns

In [25]:
# add the count to the neighborhoods
neighborhoods = neighborhoods.merge(collision_counts, how='left', on='name')
In [26]:
neighborhoods.sample(10)
Out[26]:
OBJECTID name geometry collision_counts
4 5 Baldwin Hills/Crenshaw POLYGON ((-118.36102 34.02532, -118.36026 34.0... 105
110 111 Windsor Square POLYGON ((-118.31404 34.06254, -118.31420 34.0... 25
78 79 Reseda POLYGON ((-118.55822 34.18657, -118.56223 34.1... 236
95 96 Valley Village POLYGON ((-118.37928 34.15687, -118.37769 34.1... 49
35 36 Gramercy Park POLYGON ((-118.30041 33.95002, -118.30073 33.9... 66
24 25 Eagle Rock POLYGON ((-118.19820 34.15160, -118.19821 34.1... 44
99 100 Vermont-Slauson POLYGON ((-118.28061 33.99330, -118.28061 33.9... 226
90 91 Tarzana POLYGON ((-118.52729 34.17083, -118.52724 34.1... 110
60 61 Mid-City MULTIPOLYGON (((-118.36860 34.03497, -118.3685... 212
59 60 Mar Vista POLYGON ((-118.42142 34.01033, -118.42010 34.0... 49
In [28]:
#renaming columns
collision_counts = collision_counts.rename(columns={'name': 'LA Neighborhood', 'collision_counts': 'Number of Collisions'})

Bar Chart with Top 10 Neighborhoods with the highest number of collisions¶

this data is useful to clearly see the staggering amount of collisions in Downtown LA. We can use this information to narrow down our focus to analyzing the Downtown area and see if there has been efforts from Vision Zero to reduce collisions.

In [29]:
#do a bar chart to see neighborhoods with most accidents
collision_counts[:10].plot.bar(figsize=(10,4),
                             x='LA Neighborhood', 
                             y='Number of Collisions')
Out[29]:
<AxesSubplot: xlabel='LA Neighborhood'>

Next Steps¶

I want to compare 2015 collision data to 2022 collision data and see if there has been an increase or decrease in collisions.

In [48]:
#only look at 2015 collision data
initial_year_results = client.get("d5tf-ez2w", 
                     limit = 60000, # putting an arbitrary high number (otherwise defaults to 1000)
                     where = "date_occ between '2015-01-01T00:00:00' and '2015-12-31T00:00:00'"
                    )
In [49]:
# Convert to pandas DataFrame
initial_year_results = pd.DataFrame.from_records(initial_year_results)
In [50]:
# how many rows and columns?
initial_year_results.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 52488 entries, 0 to 52487
Data columns (total 18 columns):
 #   Column        Non-Null Count  Dtype 
---  ------        --------------  ----- 
 0   dr_no         52488 non-null  object
 1   date_rptd     52488 non-null  object
 2   date_occ      52488 non-null  object
 3   time_occ      52488 non-null  object
 4   area          52488 non-null  object
 5   area_name     52488 non-null  object
 6   rpt_dist_no   52488 non-null  object
 7   crm_cd        52488 non-null  object
 8   crm_cd_desc   52488 non-null  object
 9   mocodes       51549 non-null  object
 10  vict_sex      50994 non-null  object
 11  vict_descent  50932 non-null  object
 12  premis_cd     52487 non-null  object
 13  premis_desc   52487 non-null  object
 14  location      52488 non-null  object
 15  cross_street  50764 non-null  object
 16  location_1    52488 non-null  object
 17  vict_age      43067 non-null  object
dtypes: object(18)
memory usage: 7.2+ MB
In [ ]:
In [51]:
# create latitudue/longitude columns
initial_year_results['latitude'] = pd.json_normalize(initial_year_results['location_1'])['latitude']
initial_year_results['longitude'] = pd.json_normalize(initial_year_results['location_1'])['longitude']

# convert to floats
initial_year_results['latitude'] = initial_year_results['latitude'].astype('float')
initial_year_results['longitude'] = initial_year_results['longitude'].astype('float')
In [52]:
initial_year_results.tail(5)
Out[52]:
dr_no date_rptd date_occ time_occ area area_name rpt_dist_no crm_cd crm_cd_desc mocodes vict_sex vict_descent premis_cd premis_desc location cross_street location_1 vict_age latitude longitude
52483 162004218 2015-12-31T00:00:00.000 2015-12-31T00:00:00.000 2030 20 Olympic 2076 997 TRAFFIC COLLISION 3004 3028 3030 3035 3037 3101 3401 3701 4020 M K 101 STREET VERMONT AV VENICE BL {'latitude': '34.0443', 'longitude': '-118.291... 55 34.0443 -118.2916
52484 162004058 2016-01-02T00:00:00.000 2015-12-31T00:00:00.000 1930 20 Olympic 2014 997 TRAFFIC COLLISION 3006 3028 3030 3035 3037 3101 3401 3701 4027 F W 101 STREET KINGSLEY DR 3RD ST {'latitude': '34.069', 'longitude': '-118.3029... 27 34.0690 -118.3029
52485 162004219 2015-12-31T00:00:00.000 2015-12-31T00:00:00.000 1800 20 Olympic 2013 997 TRAFFIC COLLISION 3006 3028 3030 3035 3036 3101 3401 3701 4020 F K 101 STREET WESTERN AV 1ST ST {'latitude': '34.0731', 'longitude': '-118.309... 47 34.0731 -118.3091
52486 162104010 2016-01-01T00:00:00.000 2015-12-31T00:00:00.000 1855 21 Topanga 2187 997 TRAFFIC COLLISION 3006 3037 3030 3028 3101 3401 3701 4021 F W 101 STREET VENTURA BL DON PIO AV {'latitude': '34.1685', 'longitude': '-118.605... 53 34.1685 -118.6058
52487 162104142 2015-12-31T00:00:00.000 2015-12-31T00:00:00.000 1740 21 Topanga 2185 997 TRAFFIC COLLISION 3030 3037 3028 3101 3401 3701 4021 3006 3032 M O 101 STREET TOPANGA CANYON BL DUMETZ RD {'latitude': '34.1569', 'longitude': '-118.605... 44 34.1569 -118.6058

Interactive 2015 Collision by Race Map¶

We can see just from the density of the points on this map that there is a lot more collisions in 2015 compared to our 2022 data.

In [53]:
fig = px.scatter_mapbox(initial_year_results, 
                        lat="latitude", 
                        lon="longitude", 
                        color="vict_descent",
                        labels={'vict_descent':'Race'},
                        mapbox_style="carto-positron"
                       )

# options on the layout
fig.update_layout(
    width = 800,
    height = 800,
    title = "Interactive LA Traffic Collisions Map by Race(2015)",
    title_x=0.5 # aligns title to center
)


fig.show()

get rid of null values¶

In [67]:
# get a list of distinct race values
race_list_two = initial_year_results.vict_descent.unique().tolist()
race_list_two
Out[67]:
['H',
 'X',
 'O',
 'A',
 'W',
 nan,
 'B',
 'C',
 'K',
 'F',
 'P',
 'I',
 'V',
 'J',
 'G',
 'U',
 'Z',
 'S',
 'D',
 'L',
 '-']
In [68]:
#dropping nan or null value
initial_year_results.dropna(inplace=True)

#show updated race list
race_list_two = initial_year_results.vict_descent.unique().tolist()
race_list_two
Out[68]:
['O',
 'H',
 'A',
 'W',
 'B',
 'C',
 'X',
 'K',
 'F',
 'P',
 'I',
 'V',
 'J',
 'G',
 'U',
 'Z',
 'S',
 'D',
 'L']

now lets summarize by neighborhood¶

In [69]:
# convert initial_year_results to geodataframe
gdf_two = gpd.GeoDataFrame(
    initial_year_results, geometry=gpd.points_from_xy(initial_year_results.longitude, initial_year_results.latitude),crs=4326)

# check that it can be mapped
gdf_two.plot()
Out[69]:
<AxesSubplot: >
In [70]:
# conduct a spatial join (i.e. assign each collision to the neighborhood it is in)
collision_by_neighborhood_two = gpd.sjoin(neighborhoods, gdf_two) #Spatial join Points to polygons
collision_by_neighborhood_two
Out[70]:
OBJECTID name geometry collision_counts index_right dr_no date_rptd date_occ time_occ area ... vict_sex vict_descent premis_cd premis_desc location cross_street location_1 vict_age latitude longitude
0 1 Adams-Normandie POLYGON ((-118.30069 34.03731, -118.30388 34.0... 40 49293 150326348 2015-12-11T00:00:00.000 2015-12-10T00:00:00.000 2215 03 ... M B 101 STREET VERMONT AV 25TH ST {'latitude': '34.0307', 'longitude': '-118.292... 56 34.0307 -118.2923
0 1 Adams-Normandie POLYGON ((-118.30069 34.03731, -118.30388 34.0... 40 41591 150322938 2015-10-20T00:00:00.000 2015-10-20T00:00:00.000 0815 03 ... M W 101 STREET VERMONT AV 30TH ST {'latitude': '34.0307', 'longitude': '-118.292... 37 34.0307 -118.2923
0 1 Adams-Normandie POLYGON ((-118.30069 34.03731, -118.30388 34.0... 40 29916 150318622 2015-08-20T00:00:00.000 2015-08-04T00:00:00.000 1750 03 ... F W 101 STREET LA CIENEGA BL RODEO RD {'latitude': '34.0307', 'longitude': '-118.292... 71 34.0307 -118.2923
0 1 Adams-Normandie POLYGON ((-118.30069 34.03731, -118.30388 34.0... 40 10479 150308760 2015-03-19T00:00:00.000 2015-03-18T00:00:00.000 0800 03 ... F H 101 STREET CATALINA 30TH {'latitude': '34.0272', 'longitude': '-118.293... 41 34.0272 -118.2937
0 1 Adams-Normandie POLYGON ((-118.30069 34.03731, -118.30388 34.0... 40 34018 150319482 2015-09-01T00:00:00.000 2015-08-31T00:00:00.000 2125 03 ... F H 101 STREET BUDLONG AV 29TH ST {'latitude': '34.0287', 'longitude': '-118.295... 25 34.0287 -118.2959
... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...
113 114 Lake Balboa POLYGON ((-118.50116 34.22114, -118.50116 34.2... 87 52138 151019288 2015-12-29T00:00:00.000 2015-12-29T00:00:00.000 1640 10 ... F H 101 STREET ROSCOE BL LOUISE AV {'latitude': '34.2211', 'longitude': '-118.509... 63 34.2211 -118.5099
113 114 Lake Balboa POLYGON ((-118.50116 34.22114, -118.50116 34.2... 87 47973 151018149 2015-12-01T00:00:00.000 2015-12-01T00:00:00.000 1650 10 ... F H 101 STREET ROSCOE BL LOUISE AV {'latitude': '34.2211', 'longitude': '-118.509... 44 34.2211 -118.5099
113 114 Lake Balboa POLYGON ((-118.50116 34.22114, -118.50116 34.2... 87 9853 151707385 2015-03-13T00:00:00.000 2015-03-13T00:00:00.000 2135 17 ... M O 101 STREET ROSCOE BL LOUISE AV {'latitude': '34.2211', 'longitude': '-118.509... 60 34.2211 -118.5099
113 114 Lake Balboa POLYGON ((-118.50116 34.22114, -118.50116 34.2... 87 46398 151017771 2015-11-20T00:00:00.000 2015-11-20T00:00:00.000 1145 10 ... F H 101 STREET ROSCOE BL LOUISE AV {'latitude': '34.2211', 'longitude': '-118.509... 50 34.2211 -118.5099
113 114 Lake Balboa POLYGON ((-118.50116 34.22114, -118.50116 34.2... 87 8225 151006460 2015-03-03T00:00:00.000 2015-03-02T00:00:00.000 0920 10 ... F W 101 STREET ROSCOE BL LOUISE AV {'latitude': '34.2211', 'longitude': '-118.509... 35 34.2211 -118.5099

40492 rows × 25 columns

In [56]:
# count the number of records per community
collision_counts_two = collision_by_neighborhood_two.value_counts('name').reset_index(name='collision_counts')
collision_counts_two
Out[56]:
name collision_counts
0 Downtown 2334
1 Van Nuys 1683
2 Hollywood 1649
3 Koreatown 1546
4 Westlake 1263
... ... ...
109 Elysian Valley 64
110 Cheviot Hills 62
111 Beverlywood 34
112 Hansen Dam 23
113 Chatsworth Reservoir 8

114 rows × 2 columns

In [71]:
# add the count to the neighborhoods
neighborhoods_two = neighborhoods.merge(collision_counts_two, how='left', on='name')
In [72]:
neighborhoods_two.sample(10)
Out[72]:
OBJECTID name geometry collision_counts_x collision_counts_y
11 12 Broadway-Manchester POLYGON ((-118.26954 33.92919, -118.26956 33.9... 126 342
9 10 Boyle Heights POLYGON ((-118.19263 34.06077, -118.19260 34.0... 327 1080
7 8 Beverly Grove POLYGON ((-118.37251 34.09026, -118.37250 34.0... 111 454
56 57 Lincoln Heights POLYGON ((-118.21218 34.08389, -118.21194 34.0... 107 377
32 33 Fairfax POLYGON ((-118.34563 34.06889, -118.34615 34.0... 78 394
15 16 Century City POLYGON ((-118.41267 34.06073, -118.41222 34.0... 18 103
106 107 Westlake POLYGON ((-118.25908 34.05389, -118.25984 34.0... 391 1263
36 37 Granada Hills POLYGON ((-118.48954 34.31469, -118.48576 34.3... 120 481
3 4 Atwater Village MULTIPOLYGON (((-118.27886 34.15321, -118.2788... 27 162
97 98 Venice MULTIPOLYGON (((-118.45619 33.96247, -118.4573... 150 670
In [74]:
#renaming columns
collision_counts_two = collision_counts_two.rename(columns={'name': 'LA Neighborhood', 'collision_counts': 'Number of Collisions'})
In [75]:
#do a bar chart to see neighborhoods with most accidents
collision_counts_two[:10].plot.bar(figsize=(10,4),
                             x='LA Neighborhood', 
                             y='Number of Collisions')
Out[75]:
<AxesSubplot: xlabel='LA Neighborhood'>

Interpreting 2015 vs 2022 data¶

We can see that traffic collisions have decreased dramatically from the highest rate of collisions at 2334 in 2015 to 811 collisions in 2022. We can also see that the top 10 nieghborhoods with the highest collisions have changed slightly as well. The top 3 neighborhoods has remained the same, while Koreatown and Westlake and swapped 4th and 5th place and 6th-10th place spots have changed more dramatically, with neighborhoods like Northride, East Hollywood, and Sherman Oaks falling off the list entirely in 2022. This is useful for our analysis in many ways. As we move forward, we can look in the areas that have dropped off the top 10 list and see if Vision Zero initiatives have been utilized to reduce traffic collisions. We can also look at the top 3 neighborhoods and see what has changed to make the collisions drop so dramatically.

In [ ]: